Synonyms Detection in Folk Tag Set: A Novel Hybrid Solution

Collaborative tagging is one of the most important applications of web 2.0 that allow users to associate tags (a free-form text chosen by the users) with the resource, which is metadata for that resource. These tags are used later on for search and retrieval of these resources. One of the issues in a folk tag set is ambiguity, as ambiguity causes incorrect resource(s) retrieval. To bring precision in search, we need to remove this ambiguity. One of the reasons of ambiguity is presence of synonyms in a tag set. In this work, we have proposed a novel solution for synonyms detection. The proposed solution provides a concise tagset that will be associated with the resource. The methodology of our approach can be defined in four major steps. First, we have removed misspelled tags. In the second step, we have detected synonyms using WordNet and Microsoft Word dictionaries. In the third step, we have used Euclidian distance to find rest of the synonyms and finally, we obtained precise tag set without synonyms. Dictionaries provide coverage to tags which are Standard English language words and mathematical formula covers the tags which are from folk vocabulary and are not present in the dictionaries. We have tested our approach on image resources with which tag set composed of twenty tags is associated. We compared our results with five state-of-the art techniques including cosine, Jaccard, projection, mutual information, and dice. We can conclude that the results of our approach are more accurate in finding synonyms.